We evaluated the effect of site of primary care (PC) on pattern of health care resource use in the first 3 months of life (Primary care visits, Emergency room (ER) use) by healthy newborns born at an inner city university hospital who chose either a community based clinic (CBC N=799) or a hospital based clinic (HBC n=601) for primary care. Methods: Identification of primary care site identified by the mother at the time of discharge was done by computer search and chart review. Computer and manual searches were done at each specific clinic site (CBC and HBC) and at the single children's hospital to determine PC and ER use. Data were analyzed using bivariable (T test, chi-sq) and multivariable analysis (Logistic regression). Maternal variables included age, race, prenatal care use and parity. Infant variables included birthweight and gestational age. Results: 82% CBC users vs. 73% of HBC users had at least one PC visit (P<.001 chi-sq). 30% of CBC users vs. 40% of HBC users had atleast 1 ER visit during the first 3 months of life (P<.001 chi-sq). Age at first PC visit was similar for CBC and HBC users (Mean 18 days, SD 12). HBC users were more likely to be of nonwhite race(P<.001 chi-sq). Adjusting for all maternal and infant variables, site of primary care (HBC or CBC) remained a significant predictor of health care resource use with HBC less likely to use PC (OR= 0.523, CI 0.45, 0.60, P<.0001) and more likely to use the emergency room (OR= 1.29, CI 1.14, 1.47, P<.05). Conclusion: Assuming missed opportunities for preventive care, hospital based clinics are a less optimal site for provision of primary care than community based clinics.